THUDM
Will this work well if I were to give it input sources and ask it to generate an article/report from that?
Feature request / 功能建议
Hi Thanks for a great model. I was wondering if this fits my usecase. Essentially I'm looking for grounded generation. For eg. I give it an input of some 20 documents totalling 20,000 words, and ask it to generate a long article using the documents as reference?
Motivation / 动机
A lot of times, especially in corporate settings, freeform writing is not what is needed. Rather, a report/article based on some documents/corpus is required.
Your contribution / 您的贡献
NA
Currently our LongWriter-6k training data mainly contains "Short Input+Long Output" type. The model may not generalize well to "Long Input+Long Output" type of task as your demand.
We are working on "Long Input+Long Output" tasks, such as grounded long-form generation. Stay tuned!